Demand-side refined modeling park integrated energy system

Abstract

The invention discloses a park integrated energy system for demand side refined modeling, and relates to the technical field of integrated energy systems, comprising: S1, building an overall framework of a residential park integrated energy system; s2, demand side refined modeling is carried out; s3, establishing a resident park IES operation optimization model; and S4, constructing and solving a hybrid robust optimization model. According to the demand side refined modeling park integrated energy system, the resident park IES operation optimization problem is researched, and a resident park robust operation optimization model considering demand side refined modeling is provided. Meanwhile, in order to process the uncertainty of wind and light output, a hybrid robust optimization model is provided, and a random dichotomy and Camp are combined; the CG algorithm designs a double-layer solving algorithm to solve the CG algorithm. Example results show that the DR plan of the residential park can flexibly adjust the heat / cold load of the user by adjusting the indoor temperature and the hot water temperature under the condition of keeping the energy consumption comfort of the user and the total energy consumption unchanged.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated energy systems, and particularly to a park integrated energy system with refined demand-side modeling. Background Art

[0002] An integrated energy system, namely Integrated Energy System, abbreviated as IES, is a system that integrates and synergistically utilizes multiple energy forms (such as electricity, gas, and heat energy, etc.). It realizes the hierarchical utilization and efficient economic operation of energy by integrating the advantageous characteristics of different energy forms. The IES can include links such as energy production, conversion, storage, and distribution, and interact with energy consumers to provide comprehensive energy services. In the context of IES operation optimization, a key factor in solving energy challenges is to recognize the synergy of demand response (DR) programs and the complexity introduced by renewable energy generation (REG). DR programs can enhance the flexibility of IES operation, thereby promoting the system to operate more efficiently. In addition, solving the inherent uncertainty of REG is crucial for ensuring the robustness and reliability of IES operation. Ignoring these key factors may endanger the practicality of research conclusions in IES operation.

[0003] DR makes full use of the complementary nature between multiple energy carriers in the IES to improve the economy and reliability of IES operation while ensuring the comfort of users' energy consumption. In recent years, scholars at home and abroad have conducted a large number of studies on the flexible resource modeling of the demand side of IES. Although the potential synergistic effect of DR on IES operation has been studied, the threat posed by the uncertainty of REG to the reliability of IES operation is often ignored. Therefore, the operation schemes obtained from the above studies are not truly globally optimal solutions, and may even be infeasible and unable to ensure the safety of system operation.

[0004] In recent years, although some studies have also simultaneously considered the synergistic effect of DR and the impact of REG uncertainty on IES operation. However, the DR programs in these studies are not refined enough, and the load characteristics of specific park IES have not been deeply explored.

[0005] In addition, in the past few decades, scholars at home and abroad have also studied the operation optimization methods of IES containing uncertainties. According to the modeling methods of uncertain parameters, the uncertain optimization methods are mainly divided into three types: fuzzy optimization (FO), stochastic optimization (SO), and robust optimization (RO). However, none of these three methods can limit the performance of the obtained objectives. Different from the FO, SO, and RO methods, in the field of IES operation optimization, although the IGDT method quantifies and analyzes uncertain parameters under a limited budget cost, it ignores the impact of uncertain parameters on the constraint conditions and only focuses on the performance of the objective function, unable to guarantee the safety of system operation. Even if some studies involve the safety of IES operation, the worst-case scenario is often monotonic for the constraint conditions, and the robust model is single-stage, unable to flexibly adjust the robustness of the model. Therefore, how to ensure the safety of IES operation, reduce the conservatism of decision-making, and pursue the optimality of the objective under the condition that the decision-maker's budget is limited will be the research focus of uncertain optimization methods.

[0006] In summary, the present invention is to research and improve the existing technologies and deficiencies, and provide a comprehensive energy system for a park with refined demand-side modeling. Summary of the Invention

[0007] The purpose of the present invention is to provide a comprehensive energy system for a park with refined demand-side modeling to solve the problems raised in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: A comprehensive energy system for a park with refined demand-side modeling, including:

[0009] S1. Overall framework of the comprehensive energy system for a residential park:

[0010] The comprehensive energy system for the park is specifically a residential park IES including a residential area, and the park includes heterogeneous energies such as electricity, heat energy, cold energy, and natural gas;

[0011] S2. Refined demand-side modeling:

[0012] S21. Heat load modeling: Deeply analyze the heat load characteristics of the residential park and establish a refined heat load model;

[0013] S22. Household appliance load modeling: In the residential park, each household appliance device has an operation window pre-specified by the residents; if the appliance operates within its operation window, the energy consumption comfort of the residents is satisfied.

[0014] S3. IES Operation Optimization Model for Residential Park

[0015] S31. Objective Function: Taking the structure of the integrated energy system of the residential park as the research object, an IES operation optimization model with refined modeling on the demand side is constructed.

[0016] S32. Constraint Conditions: The operation optimization model includes the operation constraints of energy equipment, such as the operation characteristic constraints of CHP units, GB, EB, EC, AC, and energy storage equipment (BS, HS, and CS), the system's power purchase and sale constraints, and the power balance constraints between power supply and demand.

[0017] S33. Compact Form of the Model: The compact form of the model to be solved is given by formulas.

[0018] S4. Construction and Solution of the Hybrid Robust Optimization Model

[0019] S41. Hybrid Robust Optimization Model: For the deterministic model formula, the GUROBI solver is used to solve it, and the optimality of the operation plan depends on the prediction accuracy of the wind and solar power output. In view of this, the HRO method is proposed to handle the uncertainty of the wind and solar power output. It can maximize the fluctuation level of the uncertain parameters while meeting the preset target performance of the decision maker, and the conservatism of the model can be controlled by the uncertainty budget Γ.

[0020] S42. Model Solution: The lower-layer two-stage robust model of the HRO model aims to find the optimal operation plan of the uncertain variable u under the worst-case scenario. For this two-stage robust optimization problem, the Column-and-Constraint Generation (C&CG) algorithm is specifically used to solve it. By decomposing the original problem into a master problem and a sub-problem, and then alternately solving them to obtain the optimal solution of the original problem.

[0021] Further, in the above S1, the internal energy conversion devices of the residential park IES include, but are not limited to, wind generators (WG), photovoltaic power generation systems (PV), combined heat and power (CHP), electric boilers (EB), gas boilers (GB), electric chillers (EC), absorption chillers (AC), battery storage (BS), cooling storage (CS), and heat storage (HS) devices. These energy conversion devices achieve the stepped utilization of energy and higher economic benefits by adjusting the transmission between various heterogeneous energies. In addition, the system is connected to the superior power grid and natural gas network through tie lines. When the on-site wind and solar power generation are insufficient, natural gas and electric energy are purchased from the external main grid to meet the energy demand.

[0022] Further, in the above S21, there are two types of heat loads in the residential park IES containing residential areas: space heat load for maintaining a comfortable room temperature and hot water load for shower and washing needs. Specific demand situation: In summer, cold energy is required for space heat demand, while heat energy is required for hot water demand. In winter, both of these heat demands consume heat energy.

[0023] 1) Space heat load modeling

[0024] The space heat load mainly comes from compensating for heat loss to maintain a suitable room temperature, and the calculation formula for the space heat load is:

[0025]

[0026] In the formula, represents the total space heat load; includes heat transfer from various building envelopes, such as doors, walls, floors, etc.; is the heat energy loss caused by wind blowing and building height; refers to the heat energy required for cold air (in winter) or hot air (in summer) leaking in from the gaps of doors and windows; describes the heat energy required to compensate for the cold air (in winter) or hot air (in summer) entering accidentally through doors or windows; is the heat energy obtained by solar irradiation on the building wall; is the heat demand generated by random and uncertain human behaviors, and is a heat load in summer and represents the generated heat energy in winter;

[0027] And The calculation formula of is as follows:

[0028]

[0029]

[0030] In the formula, α BH / δ CN / γ it is the temperature difference correction coefficient for the building envelope / crack / infiltration heat loss; K HT is the heat transfer coefficient of the room; A BH / A BW is the building envelope / south wall area; T t OT / T t IN is the outdoor / indoor temperature; δ cx / δ jx is the correlation coefficient of wind / bluilding height; I LE is the crack length of the door or window; V CL is the air volume infiltrating into the room per meter; η BI is the absorption efficiency of the wall to solar radiation; is the solar irradiance;

[0031] Modeling of thermal load demand response based on thermal inertia: Taking summer as an example, according to the first law of thermodynamics, the spatial thermal load demand can be represented by a dynamic model of temperature change, as shown in the following formula:

[0032]

[0033] In the formula, C AR is the specific heat capacity of air; ρ AR is the air density; V RB is the room volume; is the cooling energy output by the electric chiller; is the cooling energy output by the absorption chiller; P t CS,ch and P t CS,dis are the charging and discharging powers of the cold energy storage device;

[0034] Due to the existence of thermal inertia, the indoor temperature changes slowly. Therefore, within each time period (such as 1 hour), the change in indoor temperature can be ignored. Based on this, the thermal dynamics model in the formula can be converted into a state model at the hourly level, as shown in the following formula:

[0035]

[0036] 2) Hot water load modeling

[0037] For the hot water tank, it is assumed that it is always full, that is, the consumed hot water will be replaced by the same volume of cold water. Given the volume of cold water, the hot water demand can be calculated according to Equation 9; according to the first law of thermodynamics, the temperature of the water tank can be calculated by the dynamic equation Equation 10.

[0038]

[0039] In the formula, C WA is the specific heat capacity of water; ρ WA is the density of water; V t WA is the required volume of cold water; T t HW is the hot water temperature in the water tank; T t CW is the cold water temperature; and are the thermal energies generated by CHP, GB, and EB respectively; is the thermal energy consumed by AC; P t HS,ch and P t HS,dis are the charging and discharging powers of the heat storage device respectively.

[0040] Furthermore, the household appliances are divided into the following two categories:

[0041] Type I: The basic electrical load includes fixed electrical loads and random electrical loads, such as fixed electrical loads operated by residents like lights and refrigerators, and electrical demands consumed by human random behaviors, such as power-consuming devices like personal computers, TVs, and hair dryers;

[0042] Type II: Flexible electrical loads refer to some appliances that can actively participate in the DR (demand response) program under the intelligent electricity meter and financial incentives. By adjusting the operation of these household appliances, the total energy bill can be reduced;

[0043] For appliances that generate non-interruptible and deferrable loads (some appliances can transfer the electrical energy demand from high electricity price periods to low electricity price periods within a day. And once the appliance is started, it must be completed without interruption. This type of appliances includes dishwashers, dryers, washing machines, etc.), their operation constraints are described by the following formula:

[0044] indicates that the appliance *will not operate outside its operation window;

[0045] Since the electrical energy consumption of all appliances is usually discrete, during operation, they will be at the rated power level shown in this formula;

[0046] It means that the electrical appliance task should be completed within the operation window;

[0047] Furthermore, it is enforced that once the electrical appliance is started, its corresponding task will be completed without interruption;

[0048] where α * and β * are the start and stop times of the electrical appliance * respectively; E * and N * are the total electricity demand and the energy consumption duration of the electrical appliance * respectively; is the start-stop state of the electrical appliance *; is the rated power of the electrical appliance *; P t * is the output power of the electrical appliance *;

[0049] In addition, there are some interdependencies between different electrical appliances. For example, although the operation of the dryer can be delayed, it cannot be operated until the washing machine has completed its task. Therefore, in addition to the above constraint formulas 11 - 14, the dryer should also be constrained as follows:

[0050] Case 1: The dryer is used immediately after the washing machine has completed its task, and the operation constraint formula is as follows:

[0051]

[0052] where "×" means that the household electrical appliance device (e.g., dryer) operates after the household electrical appliance device * (e.g., washing machine);

[0053] Case 2: The dryer can be used after the washing machine has completed its task, as long as it is within its operation window, and the operation constraint formula is as follows:

[0054]

[0055] Furthermore, in S31, the total operating cost of the residential park mainly consists of the system's electricity purchase cost, the revenue from selling electricity to the power grid, the gas purchase cost, and the equipment maintenance cost;

[0056] min f = (C grid + C gas + C om )Δt -- Formula 17;

[0057]

[0058]

[0059] where C grid, C gas and C om are the transaction costs between the park and the main power grid, the transaction costs with the natural gas network, and the system maintenance cost respectively; T is the operation and scheduling cycle (24 hours); N and n are the total number of system energy devices and the device number respectively; C gas and C om are the unit power purchase price, power selling price, natural gas price of the residential park at time t and the unit maintenance cost of device n respectively; P t ele,buy , P t ele,sell , and P t n are the power purchase volume, power selling volume, CHP gas purchase volume, GB gas purchase volume of the system at time t and the output value of energy device n respectively.

[0060] Furthermore, in the above S32, the constraint conditions specifically include:

[0061] 1) Electric power supply-demand balance constraint

[0062] P t ele,buy +P t W +P t S +P t CHP +P t BS,dis =P t ele,sell +P t DE +P t EB +P t BS,ch +P t EC -- Formula 21;

[0063] In the formula, P t EB and P t EC are the electric powers consumed by EB and EC respectively; P t W and P t S are the outputs of wind power and photovoltaic power at time t respectively; P t BS,dis and P t BS,ch are the charge / discharge powers of the electric energy storage respectively; P t ele,buy and P tele,sell The electricity purchase and sale powers of the system in period t respectively;

[0064] 2) System electricity purchase and sale constraints

[0065] 0 ≤ P t ele,buy ≤ P t ele,max -- Formula 22;

[0066] 0 ≤ P t ele,sell ≤ P t ele,max -- Formula 23;

[0067] P t ele,buy P t ele,sell = 0 -- Formula 24;

[0068] 3) CHP unit

[0069] CHP mainly includes a gas turbine and a waste heat boiler, which can generate electric energy and heat energy simultaneously. The natural gas is burned by the gas turbine to generate electricity, and the waste heat generated during the power generation process is used by the waste heat boiler for heating or providing the heat energy required for industrial production. The operating characteristics of CHP are shown in the following formula:

[0070]

[0071] In the formula, η e and β CHP are respectively the gas purchase power, power generation efficiency and heat production efficiency of the CHP unit;

[0072] 4) GB

[0073] GB transfers the heat energy to the heat network by burning natural gas, realizing the conversion of natural gas to heat energy. Its operating characteristics are shown in the following formula:

[0074]

[0075] In the formula, η GB and are respectively the gas purchase power, conversion efficiency and maximum output power of GB;

[0076] 5) EB

[0077] EB is a device that converts electric energy into heat energy, realizing the coupling of electric energy and heat energy. Its operating characteristics are shown in the following formula:

[0078]

[0079] Wherein, η EP and are the conversion efficiency and the maximum output thermal power of EB, respectively;

[0080] 6) EC

[0081] EC is a device that converts electrical energy into cold energy, and its operating characteristics are shown in the following formula:

[0082]

[0083] Wherein, η EC and are the conversion efficiency and the maximum output power of EC, respectively;

[0084] 7) AC

[0085] AC is a device that consumes thermal energy to generate cold energy, has a high conversion efficiency, and realizes the coupling between thermal energy and cold energy. Its operating characteristics are shown in the following formula:

[0086]

[0087] Wherein, η AC and are the conversion efficiency and the maximum output cold power of AC, respectively.

[0088] 8) Energy storage operation constraints

[0089] The relationship between the state of charge (SOC) of the energy storage device and the charge and discharge power changes as follows:

[0090]

[0091] To prevent overcharging and over-discharging of the energy storage, the variable range of its SOC is restricted;

[0092] It is stipulated that the initial and final capacities of the energy storage device remain unchanged;

[0093] Wherein, B ∈ {BS, HS, CS} is the energy storage type; λ B is the self-discharge rate of the energy storage; P t B,ch and P t B,dis are the charge and discharge powers of the energy storage B at time t, respectively; is the capacity of the energy storage B at time t; η B,ch and η B,dis are the charge and discharge efficiencies of the energy storage, respectively; E B,min and E B ,maxThey are the lower and upper limits of the energy storage capacity respectively; and are the initial and final capacity sizes of Energy Storage B; Δt is the duration of a single time period;

[0094] In addition, energy storage cannot be charged and discharged simultaneously, and the charging and discharging power of the energy storage device should be restricted. Therefore, the charging and discharging power constraints of the energy storage are added as follows:

[0095]

[0096] In the formula, P t B_ch,max and P t B_dis,max are the upper limits of the charging and discharging of the energy storage respectively;

[0097] 9) Demand response constraint

[0098] Since the human body temperature has a certain adjustable range, heating has demand response characteristics. The room temperature in the residential area and the hot water temperature in the hot water tank should be kept within a certain appropriate range, and only the total energy consumption before and after DR needs to remain unchanged.

[0099] The indoor temperature and hot water temperature after DR should be within the allowable range:

[0100]

[0101] The total load size before and after DR remains unchanged:

[0102]

[0103] In the formula, and are the lower and upper limits of the indoor temperature acceptable to residents respectively; and are the lower and upper limits of the hot water temperature in the water tank acceptable to residents respectively; and are the preset indoor temperature and hot water temperature in the water tank respectively.

[0104] Furthermore, in S33, the constructed operation optimization model of the residential park is a mixed-integer linear programming problem, and Formulas 38 - 43 give the compact form of the model to be solved;

[0105]

[0106] s.t., g(x) = 0 -- Formula 39;

[0107] q(x) ≥ 0 -- Formula 40;

[0108] Ax + By ≥ h -- Formula 41;

[0109] Ex + Fy = d -- Equation 42;

[0110]

[0111] where the vector x represents the binary and continuous variables related to the residential park and the DR model, including: U * , T IN , T HW , P BS,dis , P BS,ch , P HS,dis , P HS,ch , P CS,dis and P CS,ch ; the vector y represents all other continuous variables, including: P buy , P sell , G CHP , G GB , P EB , H AC , P EC , P W and P S . Additionally, c represents the coefficient column vector related to the energy storage operation and maintenance cost; b represents the coefficient column vector related to the variable y in the objective function; h and d represent the constant column vectors related to the constraints; the vector represents the predicted values of the PV and wind power outputs, where:

[0112]

[0113] In the above model, the equality constraint formula 39 includes formulas 1 - 6, formulas 11 - 13, 36, and 37; the inequality constraint formula 40 includes formulas 14 - 16, 34, and 35. In addition, the constraint formulas 39 and 40 also include some constraints related to the energy storage operation. The constraint formulas 41 and 42 include formulas 8, 10, the operation constraints of the energy conversion devices (CHP, GB, EB, AC, EC), and the electric power balance constraint formula 21. The constraint formula 43 indicates that in the deterministic optimization model, the light and wind power outputs are the corresponding predicted values for each time period.

[0114] Furthermore, in S41, when the light and wind power outputs take the boundary values of their allowable output intervals, the most severe scenario is obtained, that is, the worst scenario occurs at the extreme points of the polyhedron. Taking wind power as an example, it can be represented by the following set:

[0115]

[0116] where u t , and Δu tThey respectively represent the actual output, predicted output, and maximum deviation of the wind power uncertainty variable in each time period. Z is a set used to control the occurrence of uncertain extreme cases, and its elements are binary variables. Γ is the uncertainty budget, which can be used to describe the severity of the uncertainty variable and reflects the decision-maker's preference for operating risks. The larger Γ is, the more cautious the decision-maker is about the range of uncertainty changes, the stronger the conservatism of the model, and the greater the cost will be paid to cope with possible risks. On the contrary, the smaller Γ is, the more optimistic the decision-maker is about the uncertainty of wind power output, and the obtained operation plan is generally too optimistic.

[0117] Furthermore, in S41, the definition of HRO is as follows:

[0118]

[0119] In the formula, f 0 is the operating cost obtained under deterministic optimization when the wind-solar output takes the predicted value; f c is the target value preset by the decision-maker, indicating the maximum operating cost that the decision-maker can bear; δ is the target deviation factor; α is the fluctuation level of the uncertain parameter, indicating the fluctuation range of the wind-solar prediction error; is the maximum deviation of the wind-solar output in each time period; γ, v, and π are the dual variables corresponding to the constraints respectively; Obviously, the constructed HRO is a two-layer optimization model. The upper layer maximizes the wind-solar fluctuation level α, and the lower layer solves the two-stage robust optimization model shown in formula 47.

[0120]

[0121] Formula 47 optimizes three different variable sets in two stages. The first min minimizes the first-stage cost by optimizing the first-stage decision variable x. These variables can be regarded as here-and-now variables (immediate decisions) because they are not affected by any wind-solar uncertainties. These variables are optimized in the first stage and remain unchanged during the optimization of the second stage. The second min aims to minimize the second-stage cost by optimizing the second-stage decision variable y under the worst wind-solar output scenario. Since the second-stage variables are optimized during the real-time operation after the uncertainty is realized, they are called wait-and-see variables (wait-and-see decisions). The worst-case scenario of the wind-solar uncertain output is obtained by maximizing the minimization of the second-stage problem, that is, by optimizing the uncertain variable u, and the wind-solar uncertain variable takes values within their respective uncertainty sets U.

[0122] In formula 47, the decision variables in the first stage are x = {U * , T IN , T HW , P BS,dis , P BS,ch , PHS,dis , P HS,ch , P CS,dis , P CS,ch}. These variables are optimized in the first stage to cope with the worst-case scenario of the uncertain output of wind and light in the second stage. The variables in the second stage are y = {P buy , P sell , G CHP , G GB , P EB , H AC , P EC , P W , P S}, which are optimized on an intraday hourly scale after the realization of wind and light uncertainty. Note: For the convenience of programming, P W and P S are included in y, but this does not affect the final optimization result of the model.

[0123] Further, in S42, the master problem - first-stage decision:

[0124]

[0125] In the formula, l is the current iteration number; y l is the solution of the sub-problem at the l-th iteration; u l is the worst-case scenario obtained at the l-th iteration.

[0126] According to the first-stage decision variable x l * obtained by solving the master problem, the worst-case scenario u l * of the wind and light output at this time can be screened by solving the lower-layer max-min sub-problem.

[0127] The sub-problem obtained by decomposing the original problem - worst-case scenario screening:

[0128]

[0129] It can be seen from the operation optimization model that the inner model of formula 49 is linear. Therefore, according to the strong duality theory, it can be transformed into the max form, and the following dual problem can be obtained by merging with the outer max:

[0130] Dual sub-problem - single-layer max model:

[0131]

[0132] Substituting formula 45 and into formula 50, we can get:

[0133]

[0134] In the formula, (z + ) T π and (z - ) T π is a bilinear term in the form of the product of a 0-1 variable and a continuous variable. By introducing auxiliary variables and related constraints, the big M method is used to perform a linear transformation on the bilinear term. The transformed sub-problem is as follows:

[0135]

[0136] Where M is a sufficiently large positive real number, z 1 and z 2 is an auxiliary variable introduced. Compared with the lower two-stage robust model, the wind and solar fluctuation level α in Formula 52 is a known quantity, which is given by the upper model.

[0137] The present invention provides a park integrated energy system with demand-side refined modeling, which has the following beneficial effects:

[0138] This paper studies the IES operation optimization problem of residential parks and proposes a robust operation optimization model for residential parks that takes into account the refined modeling of the demand side. At the same time, in order to deal with the uncertainty of wind and solar power output, a hybrid robust optimization model is proposed, and a two-layer solution algorithm is designed by combining the random dichotomy method and the C&CG algorithm to solve it. The example results show that:

[0139] 1) The DR plan for residential parks can flexibly adjust the user's heat / cooling load by adjusting the indoor temperature and hot water temperature while keeping the user's energy comfort and total energy consumption unchanged. The existence of thermal inertia makes the heating / cooling and load not exactly the same in each period. Air and hot water can be used as heat sources to store or release energy to achieve the transfer of heat load. Flexible appliances can be flexibly adjusted within their time window to change the size of the electrical load. Both electric and thermal demand response can adjust the system operation status and reduce the operating cost of residential parks.

[0140] 2) The designed two-layer solution algorithm can effectively solve the HRO robust model and ensure the optimality of the solution.

[0141] 3) Compared with deterministic optimization, SO and IGDT, the HRO method can not only quantify the uncertainty risk of wind and solar power under limited budget cost, but also flexibly control the conservatism of the system during the daily operation phase by adjusting the size of the uncertainty budget, which verifies the flexibility and advantages of HRO. BRIEF DESCRIPTION OF THE DRAWINGS

[0142] Figure 1 A schematic diagram of the structure of a residential park integrated energy system of a park integrated energy system with refined demand-side modeling according to the present invention;

[0143] Figure 2 It is a diagram of the operation optimization results obtained by the HRO method under different preset costs of a park integrated energy system with refined demand-side modeling according to the present invention;

[0144] Figure 3 It is a diagram of the deterministic optimization results of a park integrated energy system with refined demand-side modeling according to the present invention under different models;

[0145] Figure 4 It is a diagram of the electric power operation results of a park integrated energy system with refined demand-side modeling according to the present invention;

[0146] Figure 5 It is a diagram of the heat power supply and demand operation results of a park integrated energy system with refined demand-side modeling according to the present invention;

[0147] Figure 6 It is a diagram of the cooling power supply and demand operation results of a park integrated energy system with refined demand-side modeling according to the present invention;

[0148] Figure 7 It is a supply and demand curve diagram of the space cooling load with thermal inertia of a park integrated energy system with refined demand-side modeling according to the present invention;

[0149] Figure 8 It is a supply and demand curve diagram of the hot water with thermal inertia of a park integrated energy system with refined demand-side modeling according to the present invention;

[0150] Figure 9 It is a diagram of the indoor temperature before and after the heat demand response of a park integrated energy system with refined demand-side modeling according to the present invention;

[0151] Figure 10 It is a comparison diagram of the hot water temperature before and after the heat demand response of a park integrated energy system with refined demand-side modeling according to the present invention;

[0152] Figure 11 It is a diagram of the operation decision-making results of flexible household appliances before the demand response of a park integrated energy system with refined demand-side modeling according to the present invention;

[0153] Figure 12 It is a diagram of the operation decision-making results of flexible household appliances after the demand response of a park integrated energy system with refined demand-side modeling according to the present invention;

[0154] Figure 13 It is a comparison diagram of the results of four optimization methods of a park integrated energy system with refined demand-side modeling according to the present invention;

[0155] Figure 14 It is a flowchart for solving the HRO model of a park integrated energy system with refined demand-side modeling according to the present invention;

[0156] Figure 15 It is a relevant parameter diagram for the residential park heat load modeling of the park integrated energy system with refined demand-side modeling according to the present invention;

[0157] Figure 16 It is an operating parameter diagram of flexible electrical appliances in the residential park of the park integrated energy system with refined demand-side modeling according to the present invention;

[0158] Figure 17 It is a main equipment parameter diagram of IES of the park integrated energy system with refined demand-side modeling according to the present invention;

[0159] Figure 18 It is an outdoor temperature diagram for each period in summer of the park integrated energy system with refined demand-side modeling according to the present invention;

[0160] Figure 19 It is a solar irradiation intensity diagram for each period in summer of the park integrated energy system with refined demand-side modeling according to the present invention;

[0161] Figure 20 It is a hot water demand and cold water temperature diagram for each period in summer of the park integrated energy system with refined demand-side modeling according to the present invention;

[0162] Figure 21 It is a predicted value diagram of wind power and photovoltaic power in the residential park of the park integrated energy system with refined demand-side modeling according to the present invention;

[0163] Figure 22 It is a basic electricity load and flexible electricity load diagram of the residential park of the park integrated energy system with refined demand-side modeling according to the present invention;

[0164] Figure 23 It is an electric, heat and cold load diagram of the residential park before demand response of the park integrated energy system with refined demand-side modeling according to the present invention;

[0165] Figure 24 It is a power purchase and sale price diagram of the system from the main grid of the park integrated energy system with refined demand-side modeling according to the present invention. Detailed implementation manners

[0166] The following further describes the implementation manners of the present invention in detail in conjunction with the drawings and embodiments. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0167] As Figures 1 - 24 shown, a park integrated energy system with refined demand-side modeling includes:

[0168] S1. Overall framework of the residential park integrated energy system:

[0169] The park integrated energy system is specifically a residential park IES that includes residential areas, and the park includes heterogeneous energy such as electricity, heat, cold and natural gas. In this example, the overall structure of the residential park IES is as follows: Figure 1 As shown in the figure, the internal energy conversion devices of the residential park IES include but are not limited to wind generators (WG), photovoltaic cells (PV), combined heating and power (CHP), electric boilers (EB), gas boilers (GB), electric chillers (EC), absorption chillers (AC), battery storage (BS), cooling storage (CS) and heat storage (HS). These energy conversion devices adjust the transmission between various heterogeneous energy sources to achieve the step-by-step utilization of energy and higher economic benefits. For example, when the system generates excess electricity but lacks sufficient heat energy, the electric boiler can absorb the excess electricity and convert it into heat energy or store the excess electricity through the battery energy storage device, thereby promoting the economic and efficient operation of the system.

[0170] In addition, the system is connected to the upper power grid and natural gas grid through interconnection lines. When the wind and solar power generation within the park is insufficient, the energy demand can be met by purchasing natural gas and electricity from the external main grid.

[0171] S2. Demand-side refined modeling:

[0172] S21. Heat Load Modeling

[0173] The heat load characteristics of residential parks are deeply analyzed, and a refined heat load model is established; the residential park IES including residential areas has two types of heat loads: space heat load for maintaining comfortable room temperature and hot water load for showering and washing needs; specific demand conditions: in summer, cold energy is needed for space heat demand, and heat energy is used for hot water demand; in winter, both heat demands consume heat energy; considering that the change rate of heat energy is much lower than that of electric energy, it has a larger thermal inertia, and residents can tolerate temperature changes within a certain range. Therefore, the heat demand of the residential park is flexible and dispatchable.

[0174] 1) Spatial heat load modeling

[0175] The space heat load mainly comes from the compensation of heat loss to maintain a suitable room temperature, and the calculation formula of the space heat load is:

[0176]

[0177] In the formula, represents the total space heat load; includes heat transfer from various building envelopes, such as doors, walls, floors, etc.; is the heat energy loss caused by wind and building height; refers to the heat energy required for cold air (in winter) or hot air (in summer) leaking in from outside through the gaps in doors and windows; describes the heat energy that needs to be compensated for cold air (in winter) or hot air (in summer) entering accidentally through open doors or windows; is the heat energy obtained by solar irradiation on the building wall; is the heat demand generated by random and uncertain human behavior, and is a heat load in summer and represents the generated heat energy in winter;

[0178] And The calculation formula of is as follows:

[0179]

[0180] In the formula, α BH / δ CN / γ it is the temperature difference correction coefficient for building envelope / crack / infiltration heat loss; K HT is the heat transfer coefficient of the room; A BH / A BW is the area of the building envelope / south wall; T t OT / T t IN is the outdoor / indoor temperature; δ cx / δ jx is the correlation coefficient for wind / bluilding height; I LE is the crack length of the door or window; V CL is the air volume infiltrating into the room per meter; η BI is the absorption efficiency of the wall to solar radiation; is the solar irradiance;

[0181] Modeling of heat load demand response based on thermal inertia: Taking summer as an example, according to the first law of thermodynamics, the space heat load demand can be represented by a dynamic model of temperature change, as shown in the following formula:

[0182]

[0183] In the formula, C AR is the specific heat capacity of air; ρ AR is the air density; VRB is the room volume; is the cooling energy output by the electric refrigerator; C t AC is the cooling energy output by the absorption refrigerator; P t CS,ch and P t CS,dis are the charging and discharging powers of the cold energy storage device;

[0184] Due to the existence of thermal inertia, the indoor temperature changes slowly. Therefore, within each time period (e.g., 1 hour), the change in indoor temperature can be ignored. Based on this, the thermodynamic model in the formula can be converted into a state model at the hourly level, as shown in the following formula:

[0185]

[0186] 2) Hot water load modeling

[0187] For the hot water tank, it is assumed that it is always full, that is, the consumed hot water will be replaced by the same volume of cold water. Given the volume of cold water, the hot water demand can be calculated according to formula 9; according to the first law of thermodynamics, the temperature of the water tank can be calculated by the dynamic equation formula 10.

[0188]

[0189] In the formula, C WA is the heat capacity of water; ρ WA is the density of water; V t WA is the required volume of cold water; T t HW is the hot water temperature in the water tank; T t CW cold water temperature; and are the thermal energies generated by CHP, GB, and EB respectively; is the thermal energy consumed by AC; P t HS,ch and P t HS,dis are the charging and discharging powers of the heat storage device respectively;

[0190] S22. Modeling of household appliance loads

[0191] In the residential park, each household appliance device has an operation window pre-specified by the residents. If the appliance operates within its operation window, the energy consumption comfort of the residents is satisfied. Household appliances are divided into the following two categories:

[0192] Type I: The basic electrical loads include fixed electrical loads and random electrical loads, such as fixed electrical loads operated by residents like lights and refrigerators, and electrical demands consumed by human random behaviors, such as power-consuming devices like personal computers, TVs, and hair dryers.

[0193] Type II: Flexible electrical loads refer to some electrical appliances that can actively participate in the DR (Demand Response) program under the smart meter and financial incentives. By adjusting the operation of these household appliances, the total energy bill can be reduced.

[0194] For appliances that generate non-interruptible and deferrable loads (some appliances can transfer the electrical energy demand from high electricity price periods to low electricity price periods within a day. And once the appliance is started, it must be completed without interruption. This type of appliances includes dishwashers, dryers, washing machines, etc.), their operation constraints are as follows:

[0195] Indicates that the appliance *will not operate outside its operation window;

[0196] Since the electrical energy consumption of all appliances is usually discrete, during operation, they will be at the rated power level shown in this formula;

[0197] Indicates that the appliance task should be completed within the operation window;

[0198] Further enforced, once the appliance is started, its corresponding task will be completed without interruption;

[0199] In the formula, α * and β * are the turn-on and turn-off times of the appliance * respectively; E * and N * are the total electrical energy demand and energy consumption duration of the appliance * respectively; is the start-stop state of the appliance *; is the rated power of the appliance *; P t * is the output power of the appliance *;

[0200] In addition, there are certain interdependencies between different electrical appliances. For example, although the operation of the dryer is deferrable, it cannot be operated until the washing machine completes its task. Therefore, in addition to the above constraint formulas 11 - 14, the dryer should also be constrained:

[0201] Case 1: Use the dryer immediately after the washing machine completes its task, and the operation constraint formula is as follows:

[0202]

[0203] In the formula, “×” indicates that the household electrical appliance (e.g., clothes dryer) is operated after the household electrical appliance * (e.g., washing machine);

[0204] Case 2: The dryer can be used after the washing machine completes its task, as long as it is within its operating window. The operating constraint formula is as follows:

[0205]

[0206] S3, IES operation optimization model for residential parks;

[0207] S31. Objective function

[0208] by Figure 1 The IES operation optimization model with demand-side refined modeling is constructed as the research object. The total operation cost of the residential park is mainly composed of the system electricity purchase cost, the income from selling electricity to the power grid, the gas purchase cost and the equipment maintenance cost;

[0209] min f=(C grid +C gas +C om )Δt--Formula 17;

[0210]

[0211] In the formula, C grid , C gas and C om are the transaction costs between the park and the main power grid, the transaction costs with the natural gas grid, and the system maintenance costs; T is the operation scheduling cycle (24 hours); N and n are the total number of system energy equipment and equipment number, respectively; C gas and C om are the unit electricity purchase price, electricity sales price, natural gas price and unit maintenance cost of equipment n in the residential park during period t; P t ele,buy , P t ele,sell , and P t n They are the system’s power purchase, power sales, CHP gas purchase, GB gas purchase and the output of energy equipment n during period t.

[0212] S32. Constraints

[0213] The operation optimization model includes the operation constraints of energy equipment, such as the operation characteristic constraints of CHP units, GB, EB, EC, AC, energy storage equipment (BS, HS and CS), system power purchase and sales constraints, and power supply and demand balance constraints;

[0214] 1) Electric power supply and demand balance constraints

[0215] P t ele,buy +P t W +P t S +P t CHP +P t BS,dis =P t ele,sell +P t DE +P t EB +P t BS,ch +P t EC -- Formula 21;

[0216] Wherein, P t EB and P t EC are the electric powers consumed by EB and EC respectively; P t W and P t S are the outputs of wind power and photovoltaic power in period t respectively; P t BS,dis and P t BS,ch are the charge / discharge powers of the electric energy storage respectively; P t ele,buy and P t ele,sell are the purchased / sold electric powers of the system in period t respectively.

[0217] 2) System purchased / sold electricity constraint

[0218] 0 ≤ P t ele,buy ≤ P t ele,max -- Formula 22;

[0219] 0 ≤ P t ele,sell ≤ P t ele,max -- Formula 23;

[0220] P t ele,buy P t ele,sell = 0 -- Formula 24;

[0221] 3) CHP unit

[0222] CHP mainly includes a gas turbine and a heat recovery steam generator, which can generate electric energy and heat energy simultaneously. The natural gas is burned in the gas turbine to generate electricity, and the heat recovery steam generator is used to supply heat for heating or provide the heat energy required for industrial production. The operating characteristics of CHP are shown in the following formula:

[0223]

[0224] In the formula, η e and β CHP are the gas purchase power, power generation efficiency and heat production efficiency of the CHP unit respectively.

[0225] 4) GB

[0226] GB transfers heat energy to the heat network by burning natural gas, realizing the conversion of natural gas to heat energy. Its operating characteristics are shown in the following formula:

[0227]

[0228] In the formula, η GB and are the gas purchase power, conversion efficiency and maximum output power of GB respectively.

[0229] 5) EB

[0230] EB is a device that converts electric energy into heat energy, realizing the coupling of electric energy and heat energy. Its operating characteristics are shown in the following formula:

[0231]

[0232] In the formula, η EP and are the conversion efficiency and maximum output heat power of EB respectively.

[0233] 6) EC

[0234] EC is a device that converts electric energy into cold energy. Its operating characteristics are shown in the following formula:

[0235]

[0236] In the formula, η EC and are the conversion efficiency and maximum output power of EC respectively.

[0237] 7) AC

[0238] AC is a device that consumes heat energy to generate cold energy, with a high conversion efficiency, realizing the coupling between heat energy and cold energy. Its operating characteristics are shown in the following formula:

[0239]

[0240] Where η AC and are the conversion efficiency of AC and the maximum output cooling power respectively.

[0241] 8) Energy storage operation constraints

[0242] The relationship between the state of charge (SOC) of the energy storage device and the charge / discharge power changes as follows:

[0243]

[0244] To prevent overcharging and over-discharging of the energy storage, the variable range of its SOC is restricted;

[0245] It is stipulated that the initial and final capacities of the energy storage device remain unchanged;

[0246] Where B ∈ {BS, HS, CS} is the energy storage type; λ B is the self-discharge rate of the energy storage; P t B,ch and P t B,dis are the charge / discharge powers of energy storage B at time t respectively; is the capacity of energy storage B at time t; η B,ch and η B,dis are the charge / discharge efficiencies of the energy storage respectively; E B,min and E B ,max are the lower and upper limits of the energy storage capacity respectively; and are the initial and final capacity sizes of energy storage B; Δt is the duration of a single time period, which is taken as 1 hour in this embodiment.

[0247] In addition, the energy storage cannot charge and discharge simultaneously, and the charge / discharge power of the energy storage device should be restricted to a certain extent. Therefore, the charge / discharge power constraints of the energy storage are added as follows:

[0248]

[0249] Where P t B_ch,max and P t B_dis,max are the upper limits of charging and discharging of the energy storage respectively;

[0250] 9) Demand response constraints

[0251] Due to the certain adjustable range of human body temperature, heat supply has demand response characteristics. The room temperature in the residential area and the hot water temperature in the hot water tank should be maintained within a certain appropriate range, and only the total energy consumption before and after DR needs to remain unchanged.

[0252] The indoor temperature and hot water temperature after DR should be within the allowable range:

[0253]

[0254] The total load magnitude before and after DR remains unchanged:

[0255]

[0256] In the formula, and are respectively the lower and upper limits of the indoor temperature acceptable to residents; and are respectively the lower and upper limits of the hot water temperature in the water tank acceptable to residents; and are respectively the preset indoor temperature and hot water temperature in the water tank.

[0257] S33. Compact form of the model

[0258] The operation optimization model of the residential park constructed in this embodiment is a mixed-integer linear programming problem, and the compact form of the model to be solved is given by Formulas 38 - 43.

[0259]

[0260] s.t., g(x) = 0 - - Formula 39;

[0261] q(x) ≥ 0 - - Formula 40;

[0262] Ax + By ≥ h - - Formula 41;

[0263] Ex + Fy = d - - Formula 42;

[0264]

[0265] In the formula, the vector x represents the binary variables and continuous variables related to the residential park and DR model, including: U * , T IN , T HW , P BS,dis , P BS,ch , P HS,dis , P HS,ch , P CS,dis and P CS,ch ; the vector y represents all other continuous variables, including: P buy , P sell, G CHP , G GB , P EB , H AC , P EC , P W and P S . In addition, c represents the coefficient column vector related to the energy storage operation and maintenance cost; b represents the coefficient column vector related to the variable y in the objective function; h and d represent the constant column vectors related to the constraints; the vector represents the predicted values of the photovoltaic and wind power outputs, where:

[0266]

[0267] In the above model, the equality constraint formula 39 includes formulas 1 - 6, formulas 11 - 13, 36 and 37; the inequality constraint formula 40 includes formulas 14 - 16, 34 and 35. In addition, the constraint formulas 39 and 40 also include some constraints related to the energy storage operation. The constraint formulas 41 and 42 include formulas 8, 10, the operation constraints of the energy conversion equipment (CHP, GB, EB, AC, EC), and the electric power balance constraint formula 21. The constraint formula 43 indicates that in the deterministic optimization model, the light and wind power outputs are the corresponding predicted values for each time period.

[0268] S4. Construction and solution of the hybrid robust optimization model

[0269] S41. Hybrid robust optimization model

[0270] For the deterministic model formulas 38 - 43, the GUROBI solver is used to solve them, but the optimality of the obtained operation plan depends on the prediction accuracy of the wind and light power outputs. However, the prediction accuracy of the wind and light power outputs makes it difficult to guarantee the optimality of the system operation. Therefore, the operation plan obtained through the deterministic optimization model is often too risky, and the influence of uncertainty needs to be considered in the model. In this embodiment, a polyhedral uncertainty set is used to model the uncertainty of the wind and light power outputs. In addition, from the perspective of the optimality of linear programming and the system dynamic response ability, the worst-case scenario is obtained when the wind and light power outputs take the boundary values of their allowable output intervals, that is, the worst-case scenario occurs at the extreme points of the polyhedron. Taking wind power as an example, it can be represented by the following set:

[0271]

[0272] where u t , and Δu tThey respectively represent the actual output, predicted output, and maximum deviation of the wind power uncertainty variable for each time period. Z is a set used to control the occurrence of uncertain extreme cases, and its elements are binary variables. Γ is the uncertainty budget, which can be used to describe the severity of the uncertainty variable and reflects the decision maker's preference for operating risks. The larger Γ is, the more cautious the decision maker is about the range of uncertainty changes, the stronger the conservatism of the model, and the greater the cost will be paid to cope with possible risks. On the contrary, the smaller Γ is, the more optimistic the decision maker is about the uncertainty of wind power output, and the obtained operation plan is generally too optimistic;

[0273] In view of this, this embodiment combines the IGDT and two-stage robust optimization technologies to propose a new Hybrid Robust Optimization (HRO) method to handle the uncertainty of wind and light output. It can maximize the fluctuation level of uncertain parameters while meeting the preset target performance of the decision maker, and can control the conservatism of the model through the uncertainty budget Γ. The definition of HRO is as follows:

[0274]

[0275] In the formula, f 0 is the operating cost obtained under deterministic optimization when the wind and light output takes the predicted value; f c is the target value preset by the decision maker, indicating the maximum operating cost that the decision maker can bear; δ is the target deviation factor; α is the fluctuation level of the uncertain parameter, indicating the fluctuation range of the wind and light prediction error; is the maximum deviation amount of the wind and light output for each time period; γ, v, and π are the dual variables corresponding to the constraints respectively; Obviously, the HRO constructed in this embodiment is a two-layer optimization model. The upper layer is to maximize the wind and light fluctuation level α, and the lower layer is to solve the two-stage robust optimization model shown in formula 47.

[0276]

[0277] Equation 47 optimizes three different sets of variables in two stages. The first min minimizes the first-stage cost by optimizing the first-stage decision variables x. These variables can be regarded as here-and-now variables (immediate decisions) because they are not affected by any wind and solar uncertainties. These variables are optimized in the first stage and remain unchanged during the optimization of the second stage. The second min aims to minimize the second-stage cost by optimizing the second-stage decision variables y under the worst wind and solar output scenarios. Since the second-stage variables are optimized during the real-time operation after the uncertainties are realized, they are called wait-and-see variables (wait-and-see decisions). The worst-case wind and solar uncertain output is obtained by maximizing the minimization of the second-stage problem, that is, by optimizing the uncertain variable u, and the wind and solar uncertain variables take values within their respective uncertainty sets U.

[0278] In Equation 47, the decision variables in the first stage are x = {U * , T IN , T HW , P BS,dis , P BS,ch , P HS,dis , P HS,ch , P CS,dis , P CS,ch}. These variables are optimized in the first stage to cope with the worst-case wind and solar uncertain output in the second stage. The variables in the second stage are y = {P buy , P sell , G CHP , G GB , P EB , H AC , P EC , P W , P S}, which are optimized on an intraday hourly scale after the wind and solar uncertainties are realized. Note: For the convenience of programming, P W and P S are included in y, but this does not affect the final optimization result of the model.

[0279] S42. Model Solving

[0280] The lower-layer two-stage robust model of the HRO model aims to find the optimal operation plan of the uncertain variable u under the worst-case scenario. For this two-stage robust optimization problem, this embodiment uses the Column-and-Constraint Generation (C&CG) algorithm to solve it. Similar to the Benders decomposition algorithm, the C&CG algorithm also decomposes the original problem into a master problem and a subproblem, and then solves them alternately to obtain the optimal solution of the original problem. However, different from the Benders decomposition algorithm, the C&CG algorithm needs to continuously add variables and constraints related to the subproblem during the solution process of the master problem. This strategy helps the algorithm obtain a more compact lower bound of the original objective function value, thus effectively reducing the number of algorithm iterations. This change makes the C&CG algorithm have significant advantages in solving two-stage optimization problems. The master problem is obtained by decomposing the formula.

[0281] (Master Problem - First-stage Decision):

[0282]

[0283] In the formula, l is the current iteration number; y l is the solution of the subproblem in the l-th iteration; u l is the worst-case scenario obtained in the l-th iteration.

[0284] According to the first-stage decision variable x l * obtained by solving the master problem, the worst-case scenario u of the wind and light output at this time can be screened by solving the lower-layer max-min subproblem l * .

[0285] (Subproblem Obtained by Decomposing the Original Problem - Worst-case Scenario Screening):

[0286]

[0287] It can be seen from the operation optimization model of this embodiment that the inner model of formula 49 is linear. Therefore, according to the strong duality theory, it can be transformed into the max form, and the following dual problem can be obtained by combining it with the outer max:

[0288] (Dual Subproblem - Single-layer max Model):

[0289]

[0290] Substituting formula 45 and into formula 50, we can get:

[0291]

[0292] where, (z + ) T π and (z - ) T π are both bilinear terms in the form of the product of a 0-1 variable and a continuous variable. In this embodiment, by introducing auxiliary variables and related constraints and using the big M method, the bilinear terms are linearly transformed, and the transformed sub-problem is as shown in the formula:

[0293]

[0294] where, M is a sufficiently large positive real number, z 1 and z 2 are the introduced auxiliary variables. Compared with the lower-layer two-stage robust model, the wind and light fluctuation level α in formula 52 is a known quantity, which is given by the upper-layer model.

[0295] After the above derivation and transformation, both the master problem formula 48 and the sub-problem formula 52 are mixed-integer linear programming problems. In this embodiment, the C&CG algorithm is used to call the GUROBI solver to solve both of them. Then, the C&CG algorithm is combined with the random bisection method to solve the HRO robust model. The detailed solution process of the HRO model is shown in Figure 14 .

[0296] Case study:

[0297] To verify the effectiveness of the proposed HRO method and the demand response plan of the residential park, a case study simulation analysis is carried out with summer as a typical case. In the experiment, the rooms of each building in the residential park face south, and the floor area is about 200 (16 * 12.5) m 2 . The wall areas facing north / south and east / west are 44.8 (16 * 2.8) m 2 and 35 (12.5 * 2.8) m 2 , respectively. The south wall and the north wall have double-glazed windows with areas of 11.2 m 2 and 8.96 m 2 , respectively. All the windows are casement windows and there are no any blinds and sunshade devices. The area of the west door is 2.1 m 2 . The residential park contains 288 households in total, and all the walls and roofs are composed of the same structural insulation boards. The specific building heat load parameters of the residential park are shown in Figure 15 , the outdoor temperature in summer is shown in Appendix Figure 18 , the solar irradiation intensity is shown in Figure 19 , and the hot water demand and cold water temperature are shown in Figure 20 . The preset indoor room and hot water temperatures in summer are 25 °C and 75 °C respectively, and their adjustable ranges are [23 °C, 27 °C] and [72 °C, 78 °C]. The detailed operating parameters of each residential flexible electrical appliance in the residential park are shown in Figure 16As shown. In addition, the predicted values of wind power generation and photovoltaic power generation are shown in Figure 21 . The basic electrical load and flexible electrical load data of the residential park are shown in Figure 22 . The basic electric, heat and cooling loads before DR are shown in Figure 23 . The natural gas purchase price is 3.45 yuan / m 3 , and the purchase and sale electricity prices of the system from / to the main grid are shown in Figure 24 . The densities of air and water are 1.29 kg / m 3 and 997 kg / m 3 respectively, and the heat capacities of air and water are 1.006 kJ / (kg·°C) and 4.2 kJ / (kg·°C) respectively. The initial capacity of the energy storage device is 0.5 times of the maximum capacity, and the operating parameters of each energy conversion device and energy storage device are shown in Figure 17 ;

[0298] 1. Analysis of the operation optimization results based on HRO

[0299] To verify the effectiveness of the proposed HRO method, Γ is set to 18, and the convergence thresholds of the upper-layer stochastic dichotomy method and the lower-layer C&CG algorithm of the solution algorithm are set to 10 -4 and 10 -6 . Then experiments are carried out under different target deviation factors δ to analyze the performance of the HRO method in dealing with the uncertainty of wind and light output. First, when not considering the volatility of wind and light output, the operating cost obtained by solving the deterministic operation optimization model of the residential park IES containing the DR plan is 12,076 yuan. Then different target deviation factors are set to obtain the maximum operating cost that the decision maker can bear. Finally, considering the DR plan, the HRO method is used to solve the operation optimization model of the residential park containing the uncertainty of wind and light output.

[0300] Figure 2 shows the operating cost, wind and light fluctuation level and solution time obtained by the HRO method under different preset costs for the decision maker. As can be seen from Figure 2 , as the preset cost of the decision maker increases, the wind and light fluctuation level obtained by HRO also increases, indicating that the system's ability to resist wind and light fluctuations during the intraday operation stage is stronger. In addition, the solution time of the HRO method under all δ is less than 5 minutes, which is completely acceptable for the decision maker to formulate the day-ahead operation plan.

[0301] In addition, the optimized operating costs obtained by the HRO method in Figure 2 are not strictly less than or equal to their preset costs under some target deviation factors, which is caused by the convergence threshold of the upper-layer stochastic dichotomy method. Since in the later stage of the solution algorithm iteration, the influence of the fluctuation level on the optimization result is not sensitive, and the gap between the optimized operating cost and the preset cost is not large. Therefore, the convergence threshold of the stochastic dichotomy method is set to 10-4 is reasonable.

[0302] 2. Validation of the effectiveness of demand response

[0303] To verify the effectiveness of the demand response plan in the residential park, in this subsection, the GUROBI solver is used to perform deterministic optimization on the optimization models of the residential park without DR, with electric DR, with thermal DR, and with electric-thermal DR, Figure 3 and list the operation optimization results under the four models. From Figure 3 it can be seen that except for the operation optimization model with electric-thermal DR, there are no feasible solutions for other models, which is sufficient to verify the effectiveness of the electric-thermal DR plan.

[0304] To verify Figure 3 the effectiveness of the operation optimization strategy obtained with DR in Figures 4 - 6 shows the optimization results corresponding to the electric-thermal-cooling energy flow. From Figures 4 - 6 it can be seen that the electric power of the residential park has met the supply-demand balance. However, due to the existence of thermal inertia, the cooling and heating powers do not need to be balanced in real time. They only need to meet the appropriate temperature range of users while keeping the total energy consumption within the operation cycle unchanged. From Figure 4 it can be seen that when the wind power output is high, the electrical load is small, and the system's electricity purchase price is low at night, the system mainly meets the residential electricity demand through wind power output and power purchase from the main grid, and at this time, the electrical energy storage is in the charging state. During the peak electricity consumption period during the day, the system is mainly powered by wind power, photovoltaic power, and CHP, and the energy storage is in the discharging state. In addition, during the peak electricity price period, the system also sells electricity to the grid to effectively reduce the operation cost. From Figure 5 it can be seen that since the thermal load is small and the cooling load is large in summer, and the electrical load is large during the day, a large amount of heat energy is generated while the CHP is supplying power. Therefore, during the day, the AC absorbs the excess heat energy and provides cooling energy, and when the thermal load is small, the thermal energy storage also stores the excess heat energy. In addition, from Figure 6 it can be seen that the cooling load is large in summer, so the AC is in the full-load state in most periods, and the cold energy storage discharges energy during the peak cooling period.

[0305] Figure 7 presents the total demand curves of cooling supply with thermal inertia and cooling load. Among them, the total cooling load includes the cooling demand of the user space and the cooling energy stored and released by the cold energy storage. In addition, Figure 8 shows the hot water supply and demand curves with thermal inertia. Among them, the hot water demand covers the hot water demand of users, the heat energy absorbed by the absorption chiller, and the heat energy stored and released by the thermal energy storage. The shaded part represents the difference between the supply and demand of thermal power and cooling power due to the existence of thermal inertia. From Figure 7 and Figure 8 it can be seen that the thermal and cooling supply and demand are not strictly equal, and there is a certain degree of adjustability.

[0306] To further verify the effectiveness of the thermal DR plan in residential parks, Figure 9 and Figure 10 respectively show the comparison results of the indoor temperature and the hot water temperature in the hot water tank before and after optimization. As can be seen from the figure, significant changes have occurred in the indoor temperature and the hot water temperature at different time periods after DR. Thermal DR changes the space cooling load and the hot water load by adjusting the indoor temperature and the hot water temperature in the hot water tank. Therefore, thermal DR can flexibly adjust the user's heating and cooling load while maintaining the user's energy consumption comfort and total energy consumption unchanged, thereby reducing the total operating cost of the system, which has a positive effect on the economic operation of the system.

[0307] Figure 11 and Figure 12 show the comparison results of the operation decision-making of household appliances before and after DR on a typical summer day. From Figure 11 and Figure 12 it can be seen that household appliances participate in flexible scheduling under DR. Compared with the preferred operation time periods of all appliances without DR in Figure 11 , household appliances participate in flexible scheduling during the time periods with lower electricity prices in Figure 12 , such as the time period from 7:00 to 15:00. Flexible appliances can be flexibly adjusted within their time windows to reduce the operating cost of residential parks.

[0308] In summary, the DR plan for residential parks can flexibly adjust the heating, cooling, and electrical loads according to the system's electricity purchase and sale prices and the user's temperature comfort range during the day-ahead operation decision-making stage, so that the operating state of the system reaches the optimal, thereby reducing the system operating cost.

[0309] 3. Comparative analysis of different methods

[0310] To verify the advantages of the operation plan obtained by the proposed HRO method, this subsection conducts experimental comparisons of HRO with deterministic optimization, SO, and IGDT methods. Among them, the preset costs of IGDT and HRO are both set to 14,492 yuan. After executing the day-ahead operation optimization program, the wind and light fluctuation levels obtained by the IGDT and HRO methods are 0.1928 and 0.2054 respectively. SO uses 10 typical scenarios to optimize and solve the system's day-ahead operation plan. To compare the advantages and disadvantages of the decision-making plans obtained by the four methods during intraday operation, in this experiment, 2000 typical random wind power and photovoltaic output scenarios are generated by the Latin hypercube sampling method within the interval of the wind and light prediction error of 0.2054. Then, the first-stage optimization variables (day-ahead decision-making plan) are fixed, and the 2000 wind and light typical scenarios are substituted into the intraday operation optimization model for solution.

[0311] Figure 13 lists the optimization and solution results of the four methods. From Figure 13It can be seen that although the model solution times of the SO and HRO methods are relatively long, the intraday solution times of all four methods are within an acceptable range. In addition, the day-ahead operating costs and scenario average operating costs obtained by the deterministic optimization and SO methods are relatively small, being relatively smaller than those of IGDT and HRO. However, their feasible rates are only 68.45% and 75.34% respectively, and neither can guarantee the safe operation of the intraday system. Although the day-ahead operating costs and scenario average operating costs obtained by the IGDT and HRO methods are relatively high, their feasible rates are much greater than those of the deterministic optimization and SO methods. In addition, HRO achieves a scenario average operating cost similar to that of IGDT, but its feasible rate is slightly higher than that of the IGDT method. Additionally, IGDT cannot adjust the robustness of the optimization decision scheme, while HRO can not only quantify the uncertainty risk of wind and light, but also flexibly control the conservatism of the system during the intraday operation stage by adjusting the magnitude of Γ, which also reflects the flexibility and advantages of the HRO method.

[0312] The embodiments of the present invention are given for purposes of illustration and description, and are not exhaustive or limit the invention to the disclosed form. Many modifications and variations are obvious to those of ordinary skill in the art. The embodiments are chosen and described in order to better explain the principles of the present invention and its practical applications, and to enable those of ordinary skill in the art to understand the present invention so as to design various embodiments with various modifications suitable for specific purposes.

Claims

1. A park integrated energy system with refined demand-side modeling, characterized in that: include: S1. Overall framework of integrated energy system in residential areas: The park integrated energy system is specifically a residential park IES including residential areas, and the park includes heterogeneous energy sources such as electricity, heat, cold and natural gas; S2. Demand-side refined modeling: S21. Heat load modeling: In-depth analysis of the heat load characteristics of residential parks and establishment of a refined heat load model; S22. Household appliance load modeling: In a residential park, each household appliance has an operating window pre-specified by the residents; if the appliance operates within its operating window, the energy comfort level of the residents is met. S3. IES operation optimization model for residential parks: S31. Objective function: Taking the integrated energy system structure of residential parks as the research object, an IES operation optimization model with refined demand-side modeling is constructed; S32. Constraints: The operation optimization model includes the operation constraints of energy equipment, system power purchase and sales constraints, and power supply and demand balance constraints; S33, compact form of model: give the compact form of the model to be solved through formula; S4. Construction and solution of hybrid robust optimization model: S41. Hybrid robust optimization model: For the deterministic model formula, the GUROBI solver is used to solve it, and it is found that the optimality of the operation scheme depends on the prediction accuracy of wind and solar power output. In view of this, the HRO method is proposed to deal with the uncertainty of wind and solar power output. S42, model solution: For the two-stage robust optimization problem, the C&CG algorithm is used to solve it, by decomposing the original problem into a main problem and a sub-problem, and then solving them alternately to obtain the optimal solution of the original problem.

2. According to claim 1, a campus integrated energy system with refined demand-side modeling is characterized in that: In S1, the internal energy conversion devices of the residential park IES include but are not limited to wind turbines, photovoltaic power generation systems, cogeneration, electric boilers, gas boilers, electric refrigerators, absorption refrigerators, electric energy storage, cold energy storage and thermal energy storage devices, and the system is connected to the upper-level power grid and natural gas grid through interconnection lines. When the wind and solar power generation within the park is insufficient, the energy demand is met by purchasing natural gas and electricity from the external main grid.

3. According to claim 1, a campus integrated energy system with refined demand-side modeling is characterized in that: In S21, there are two types of heat loads in the residential park IES: space heat load for maintaining a comfortable room temperature and hot water load for showering and washing needs; Specific demand situations: In summer, cold energy is needed for space heating demand and hot energy for hot water demand; in winter, both heat demands consume hot energy; 1) Spatial heat load modeling The space heat load mainly comes from the compensation of heat loss to maintain a suitable room temperature, and the calculation formula of the space heat load is: In the formula, represents the total space heat load; Includes heat transfer from various building envelope structures such as doors, walls, floors, etc.; is the heat loss caused by wind and building height; Refers to the heat energy required for the cold air (in winter) or hot air (in summer) that leaks in from the gaps between doors and windows. Describes the thermal energy required to compensate for the cold air (in winter) or hot air (in summer) that enters through the accidental opening of a door or window; The heat energy obtained from solar radiation on the building walls; is the heat demand generated by random given human uncertain behavior, and In summer it is heat load, in winter it is expressed as heat energy generated; and The calculation formula is as follows: In the formula, α BH / δ CN / γ it K is the temperature difference correction factor for building envelope / crack / intrusion heat loss; HT is the heat transfer coefficient of the room; A BH / A BW is the building envelope / south wall area; T t OT / T t IN is the outdoor / indoor temperature; cx / δ jx is the correlation coefficient of wind / building height; I LE V is the length of the crack in the door or window; CL is the amount of air that penetrates into the room per meter; η BI is the absorption efficiency of the wall to solar radiation; is the solar irradiance; Thermal inertia-based heat load demand response modeling: Taking summer as an example, according to the first law of thermodynamics, the space heat load demand can be expressed by a dynamic model of temperature change, as shown in the following formula: In the formula, C AR is the heat capacity of air; ρ AR is the air density; V RB is the room volume; It is the cold energy output by the electric refrigerator; P is the cooling energy output by the absorption refrigerator; t CS,ch and P t CS,dis It is the charging and discharging power of the cold energy storage equipment; Based on the existence of thermal inertia, the thermal dynamics model in the formula can be converted into an hourly state model, as shown in the following formula: 2) Hot water load modeling For the hot water tank, it is assumed that it is always full, that is, the consumed hot water will be replaced by the same volume of cold water; given the cold water volume, the hot water demand can be calculated according to formula 9; according to the first law of thermodynamics, the temperature of the water tank can be calculated by the dynamic equation formula 10; H t HW = C WA ρ WA V t WA (T t HW - T t CW ) -- Equation 9; In the formula, C WA is the heat capacity of water; ρ WA is the density of water; V t WA is the required cold water volume; T t HW is the hot water temperature in the water tank; T t CW Cold water temperature; and are the heat energy generated by CHP, GB and EB respectively; P is the heat energy consumed by AC; t HS,ch and P t HS,dis are the charging and discharging power of the heat storage equipment respectively.

4. According to claim 1, a campus integrated energy system with refined demand-side modeling is characterized in that: The household appliances are divided into the following two categories: Type I: Basic electricity load includes fixed electricity load and random electricity load, fixed electricity load operated by residents and electricity demand consumed by random human behavior; Type II: Flexible electric loads refer to some electrical appliances that can actively join the DR program with smart meters and financial incentives; For appliances generating non-interruptible and deferrable loads, the operating constraints are as described in the following formula: In the formula, α * and β * are the opening and closing times of the electrical appliance*; E * and N * are the total electricity demand and energy consumption duration of electrical appliances*; The start and stop status of the electrical appliance*; P is the rated power of the appliance*; t * The output power of the appliance*; In addition, the dryer cannot be operated before the washing machine completes its task. In addition to the above constraint formulas 11-14, the dryer should also be constrained: Case 1: Use the dryer immediately after the washing machine completes its task. The running constraint formula is as follows: Wherein, "×" indicates that the household electrical appliance (e.g., clothes dryer) is operated after the household electrical appliance * (e.g., washing machine); Case 2: The dryer can be used after the washing machine completes its task, as long as it is within its operating window. The operating constraint formula is as follows:

5. According to claim 1, a campus integrated energy system with demand-side refined modeling is characterized in that: In S31, the total operating cost of the residential park is composed of the system electricity purchase cost, the income from selling electricity to the power grid, the gas purchase cost and the equipment maintenance cost; minf=(C grid +C gas +C om )Δt--Formula 17; In the formula, C grid , C gas and C om are the transaction costs between the park and the main power grid, the transaction costs with the natural gas grid, and the system maintenance costs; T is the operation scheduling cycle; N and n are the total number of system energy equipment and equipment number, respectively; C gas and C om are the unit electricity purchase price, electricity sales price, natural gas price and unit maintenance cost of equipment n in the residential park during period t; P t ele,buy , P t ele,sell , and P t n They are the system’s power purchase, power sales, CHP gas purchase, GB gas purchase and the output of energy equipment n during period t.

6. The park integrated energy system with demand-side refined modeling according to claim 1 is characterized in that: In S32, the constraints specifically include: 1) Electric power supply and demand balance constraints P t ele,buy +P t W +P t S +P t CHP +P t BS,dis =P t ele,sell +P t DE +P t EB +P t BS,ch +P t EC --Formula 21; Where P t EB and P t EC are the electric power consumed by EB and EC respectively; P t W and P t S are the output of wind power and photovoltaic power in period t respectively; P t BS,dis and P t BS,ch are the charging / discharging power of the electric energy storage; P t ele,buy and P t ele,sell They are the purchased and sold power of the system during period t; 2) System power purchase and sales constraints 0≤P t ele,buy ≤P t ele,max --Formula 22; 0≤P t ele,sell ≤P t ele,max --Formula 23; P t ele,buy P t ele,sell =0--Formula 24; 3) CHP unit The operating characteristics of CHP are shown below: In the formula, η e and β CHP They are the gas purchase power, power generation efficiency and heat generation efficiency of the CHP unit respectively; 4GB GB operating characteristics are shown below: In the formula, η GB and They are GB's gas purchase power, conversion efficiency and maximum output power; 5)EB The operating characteristics of EB are shown below: Where η EP and are the conversion efficiency and maximum output thermal power of EB respectively; 6)EC The EC operating characteristics are shown below: Where η EC and are the conversion efficiency and maximum output power of EC respectively; 7) AC The AC operating characteristics are shown below: Where η AC and They are AC conversion efficiency and maximum output cooling power respectively; 8) Energy storage operation constraints The relationship between the energy storage state of the energy storage device and the charging and discharging power is as follows: To prevent over-charging and over-discharging of energy storage, its SOC variable range is constrained; It is stipulated that the capacity of energy storage equipment remains unchanged from beginning to end; Where, B∈{BS,HS,CS} is the energy storage type; λ B is the energy storage self-discharge rate; P t B,ch and P t B,dis are the charging and discharging power of energy storage B in period t respectively; is the capacity of energy storage B during period t; η B,ch and η B,dis are the charging and discharging efficiency of energy storage; E B,min and E B,max are the lower and upper limits of energy storage capacity respectively; and is the initial and final capacity of energy storage B; Δt is the duration of a single period; In addition, energy storage cannot be charged and discharged at the same time, and the charging and discharging power of energy storage devices should be subject to certain restrictions. The charging and discharging power constraints of added energy storage are as follows: Where P t B_ch,max and P t B_dis,max They are the upper limits of energy storage charging and discharging respectively; 9) Demand response constraints The indoor temperature and hot water temperature after DR should be within the allowable range: The total load size before and after DR remains unchanged: In the formula, and They are the lower and upper limits of indoor temperatures acceptable to residents, respectively; and They are the lower and upper limits of the water tank hot water temperature that residents can accept, respectively; and They are the preset indoor temperature and the water tank hot water temperature respectively.

7. The park integrated energy system with demand-side refined modeling according to claim 1 is characterized in that: In said S33, the residential park operation optimization model constructed is a mixed integer linear programming problem, and formulas 38-43 give the compact form of the model to be solved; st,g(x)=0--Formula 39; q(x)≥0--Formula 40; Ax+By≥h--Formula 41; Ex+Fy=d--Formula 42; In the formula, vector x represents binary variables and continuous variables related to the residential park and DR model, including: U * , T IN , T HW , P BS ,dis , P BS,ch , P HS,dis , P HS,ch , P CS,dis and P CS,ch ; vector y represents all other continuous variables including: P buy , P sell , G CHP , G GB , P EB , H AC , P EC , P W and P S ; In addition, c represents the coefficient column vector related to the energy storage operation and maintenance cost; b represents the coefficient column vector related to the variable y in the objective function; h and d represent the constant column vectors related to the constraints; vector Represents the predicted value of photovoltaic and wind power output, where:

8. The park integrated energy system with demand-side refined modeling according to claim 1 is characterized in that: In S41, when the wind and solar power output takes the boundary value of its allowable output range, the worst scenario is obtained, that is, the worst scenario occurs at the extreme point of the polyhedron. Taking wind power as an example, it can be represented by the following set: In the formula, u t , and Δu t They represent the actual output, predicted output and maximum output deviation of the uncertain variables of wind power in each period respectively; Z is a set used to control the occurrence of uncertain extreme situations, and its elements are binary variables; Γ is the uncertainty budget, which can be used to describe the severity of the uncertainty variables and reflects the decision maker's preference for operational risks.

9. The park integrated energy system with demand-side refined modeling according to claim 1 is characterized in that: In S41, HRO is defined as follows: Where f0 is the operating cost obtained under deterministic optimization when the wind and solar power output takes the predicted value; f c The target value preset for the decision maker represents the maximum operating cost that the decision maker can bear; δ is the target deviation factor; α is the fluctuation level of the uncertain parameter, indicating the fluctuation amplitude of the wind and solar forecast error; γ, v and π are the dual variables of the corresponding constraints; Obviously, the constructed HRO is a two-layer optimization model, the upper layer is to maximize the wind and solar fluctuation level α, and the lower layer is to solve the two-stage robust optimization model shown in Formula 47; In formula 47, the decision variable of the first stage is x = {U * , T IN , T HW , P BS,dis , P BS,ch , P HS,dis , P HS,ch , P CS,dis , P CS,ch }, and optimize it in the first stage; The variable of the second stage is y = {P buy , P sell , G CHP , G GB , P EB , H AC , P EC , P W , P S }, which is optimized at the intraday and hourly scale after the uncertainty of wind and solar power is realized.

10. The park integrated energy system with demand-side refined modeling according to claim 1 is characterized in that: In S42, the main problem is the first stage decision: Where l is the current iteration number; y l is the solution of the l-th iteration subproblem; u l is the worst scenario obtained in the lth iteration; Based on the first-stage decision variable x obtained by solving the main problem l * By solving the lower-level max-min subproblem, we can select the worst scenario u for wind and solar output at this time. l * ; Sub-problems obtained by decomposing the original problem - worst scenario screening: From the running optimization model, we can see that the inner model of Formula 49 is linear. Therefore, according to the strong duality theory, it can be transformed into the form of max, and by merging it with the outer max, we can get the following dual problem: Dual sub-problem - single-layer max model: Substituting Equation 45 and Substituting into formula 50, we get: In the formula, (z + ) T π and (z - ) T π is a bilinear term in the form of the product of a 0-1 variable and a continuous variable. By introducing auxiliary variables and related constraints, the big M method is used to perform a linear transformation on the bilinear term. The transformed subproblem is as follows: In the formula, M is a sufficiently large positive real number, z1 and z2 are introduced auxiliary variables; relative to the lower two-stage robust model, the wind and solar fluctuation level α in Formula 52 is a known quantity, which is given by the upper model.

Images